<?php
/**
 * URL extraction from shortcode attributes and block attributes.
 *
 * @package Unbreak
 */

namespace Unbreak\Extract;

use Unbreak\Support\Url;

defined( 'ABSPATH' ) || exit;

/**
 * Finds URLs stored outside HTML attributes: page-builder shortcode attributes (Divi, WPBakery, ...)
 * and Gutenberg block comment attributes (embeds, file blocks, ...).
 */
final class AttributeParser {

	/**
	 * Attribute names that hold a URL: url, link, href, src, or anything ending in _url/_link/_href/_src.
	 */
	private const SHORTCODE_KEY = '/(?:^|_)(?:url|link|href|src)$/i';

	/**
	 * Block attribute names that hold a URL: url, href, link, or camelCase names ending in Url/Link/Href.
	 */
	private const BLOCK_KEY = '/^(?:url|href|link|src)$|[a-z](?:Url|Link|Href|Src)$/';

	/**
	 * Sibling attributes used as anchor text.
	 */
	private const TEXT_KEYS = array( 'button_text', 'text', 'title', 'link_text', 'alt', 'label' );

	/**
	 * Shortcode attribute pass. Shortcodes are not HTML, so a tokenizer regex is appropriate here.
	 *
	 * @param string $content        Post content.
	 * @param string $base           Base URL.
	 * @param string $source_key     Source key.
	 * @param bool   $include_images Include image URLs.
	 * @return array<int,array{url:string,element:string,anchor_text:string,context:string,source_key:string}>
	 */
	public static function shortcodes( string $content, string $base, string $source_key = '', bool $include_images = true ): array {
		if ( false === strpos( $content, '[' ) ) {
			return array();
		}
		if ( ! preg_match_all( '/\[(?![\/\[])([a-zA-Z][a-zA-Z0-9_\-]*)(\s[^\]]*?)?\/?\]/', $content, $matches, PREG_SET_ORDER ) ) {
			return array();
		}

		$found = array();
		foreach ( $matches as $match ) {
			$tag      = $match[1];
			$atts_raw = isset( $match[2] ) ? trim( $match[2] ) : '';
			if ( '' === $atts_raw ) {
				continue;
			}
			$atts = shortcode_parse_atts( $atts_raw );
			if ( ! is_array( $atts ) ) {
				continue;
			}
			$anchor = '';
			foreach ( self::TEXT_KEYS as $text_key ) {
				if ( isset( $atts[ $text_key ] ) && is_string( $atts[ $text_key ] ) && '' !== trim( $atts[ $text_key ] ) ) {
					$anchor = HtmlParser::squash( wp_strip_all_tags( $atts[ $text_key ] ) );
					break;
				}
			}
			foreach ( $atts as $key => $value ) {
				if ( ! is_string( $key ) || ! is_string( $value ) || ! preg_match( self::SHORTCODE_KEY, $key ) || ! self::looks_like_url( $value ) ) {
					continue;
				}
				$element = self::is_image_key( $key ) ? 'img' : 'a';
				if ( 'img' === $element && ! $include_images ) {
					continue;
				}
				$url = Url::normalize( $value, $base );
				if ( null === $url ) {
					continue;
				}
				$found[] = array(
					'url'         => $url,
					'element'     => $element,
					'anchor_text' => HtmlParser::truncate( $anchor, 255 ),
					'context'     => '[' . $tag . ' ' . $key . ']',
					'source_key'  => $source_key,
				);
			}
		}
		return $found;
	}

	/**
	 * Block attribute pass over serialized blocks.
	 *
	 * @param string $content        Post content.
	 * @param string $base           Base URL.
	 * @param string $source_key     Source key.
	 * @param bool   $include_images Include image URLs.
	 * @return array<int,array{url:string,element:string,anchor_text:string,context:string,source_key:string}>
	 */
	public static function blocks( string $content, string $base, string $source_key = '', bool $include_images = true ): array {
		if ( false === strpos( $content, '<!-- wp:' ) || ! function_exists( 'parse_blocks' ) ) {
			return array();
		}
		$found = array();
		self::walk_blocks( parse_blocks( $content ), $base, $source_key, $include_images, $found );
		return $found;
	}

	/**
	 * Recursive helper for blocks().
	 *
	 * @param array<int,mixed>                                                                                $blocks         Parsed blocks.
	 * @param string                                                                                          $base           Base URL.
	 * @param string                                                                                          $source_key     Source key.
	 * @param bool                                                                                            $include_images Include images.
	 * @param array<int,array{url:string,element:string,anchor_text:string,context:string,source_key:string}> $found          Accumulator.
	 */
	private static function walk_blocks( array $blocks, string $base, string $source_key, bool $include_images, array &$found ): void {
		foreach ( $blocks as $block ) {
			if ( ! is_array( $block ) ) {
				continue;
			}
			$name  = isset( $block['blockName'] ) && is_string( $block['blockName'] ) ? $block['blockName'] : '';
			$attrs = isset( $block['attrs'] ) && is_array( $block['attrs'] ) ? $block['attrs'] : array();
			foreach ( $attrs as $key => $value ) {
				if ( ! is_string( $key ) || ! is_string( $value ) || ! preg_match( self::BLOCK_KEY, $key ) || ! self::looks_like_url( $value ) ) {
					continue;
				}
				$element = self::is_image_key( $key ) ? 'img' : 'a';
				if ( 'img' === $element && ! $include_images ) {
					continue;
				}
				$url = Url::normalize( $value, $base );
				if ( null === $url ) {
					continue;
				}
				$found[] = array(
					'url'         => $url,
					'element'     => $element,
					'anchor_text' => '',
					'context'     => $name . ' block',
					'source_key'  => $source_key,
				);
			}
			if ( ! empty( $block['innerBlocks'] ) && is_array( $block['innerBlocks'] ) ) {
				self::walk_blocks( $block['innerBlocks'], $base, $source_key, $include_images, $found );
			}
		}
	}

	/**
	 * Whether a value plausibly is a URL (absolute, scheme-relative or root-relative).
	 *
	 * @param string $value Value.
	 */
	public static function looks_like_url( string $value ): bool {
		$value = trim( $value );
		return (bool) preg_match( '#^(?:https?:)?//[^\s/]+|^/[^/\s]#i', $value );
	}

	/**
	 * Whether an attribute name suggests an image.
	 *
	 * @param string $key Attribute name.
	 */
	private static function is_image_key( string $key ): bool {
		return (bool) preg_match( '/src|image|img|media/i', $key );
	}
}